Presentation 2003/9/2
Face Recognition Using Pose-Free Standard Geodesic Weights
Masahiko HAMANAKA, Rui ISHIYAMA, Shizuo SAKAMOTO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) We are developing a face recognition system using Geodesic Illumination Basis Method which can realize high accuracy against pose and illumination changes by using 3D models. We propose a weighted matching method using 2D weights dependent on an estimated pose, which are generated from pose-free standard geodesic weights learned on the averaged 3D model. This method can realize face recognition under any pose based on the stability or distinctiveness of features dependent on their positions on the face. In addition, we propose how to learn weights using not only average errors from correct models but also those from wrong models. In the experiments using facial image database of 200 individuals, the proposed method reduced more than 30% of errors compared with the previous method, and achieved a first-choice success ratio of 94.9%.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) face recognition / weighted matching / pose / illumination / 3D model / average error
Paper # PRMU2003-105
Date of Issue

Conference Information
Committee PRMU
Conference Date 2003/9/2(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Pattern Recognition and Media Understanding (PRMU)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Face Recognition Using Pose-Free Standard Geodesic Weights
Sub Title (in English)
Keyword(1) face recognition
Keyword(2) weighted matching
Keyword(3) pose
Keyword(4) illumination
Keyword(5) 3D model
Keyword(6) average error
1st Author's Name Masahiko HAMANAKA
1st Author's Affiliation Multimedia Research Laboratories, NEC Corporation()
2nd Author's Name Rui ISHIYAMA
2nd Author's Affiliation Multimedia Research Laboratories, NEC Corporation
3rd Author's Name Shizuo SAKAMOTO
3rd Author's Affiliation Multimedia Research Laboratories, NEC Corporation
Date 2003/9/2
Paper # PRMU2003-105
Volume (vol) vol.103
Number (no) 296
Page pp.pp.-
#Pages 6
Date of Issue